Restart ALL pods in Kubernetes cluster

Yes, someday you may need this.

1. Do it soft way.

This one takes some time to finish. Depends on the number of POD in your cluster.

kubectl get ns -o name|awk -F/ {'print $2'} |grep -vE "^kube-system$"| xargs -I{} kubectl -n {} rollout restart deploy
kubectl get ns -o name|awk -F/ {'print $2'} |grep -vE "^kube-system$"| xargs -I{} kubectl -n {} rollout restart ds
kubectl get ns -o name|awk -F/ {'print $2'} |grep -vE "^kube-system$"| xargs -I{} kubectl -n {} rollout restart sts

2. Do it hard way.

#!/bin/bash
for i in $(kubectl get pods -A|grep -vE "^NAMESPACE|^kube-system"|shuf|awk {'print $1","$2'})
do
IFS=', ' read -r -a arr <<< "$i"
echo "NameSpace: " ${arr[0]}
echo "Pod ID: " ${arr[1]}
echo "kubectl delete pod ${arr[1]} -n ${arr[0]}"
kubectl delete pod ${arr[1]} -n ${arr[0]} &>/dev/null &
#little sleep, reduce load to api-server and scheduler
sleep 0.5
done

Good luck.